SSyncDev

pplx-decider-flash-v1.3

SyncDev model directory · Profile updated Nov 17, 2026

Compare with another model

pplx-decider-flash-v1.3 is Perplexity's small, half-price decision model, distilled from the 27B decider, for high-volume classification and span extraction.

Key specifications

Specification score
10.0
Input / 1M tokens
From $0.01
License
Apache-2.0

The specification score summarizes recorded features and limits. It is not a benchmark result.

Pricing conditions: Indicative: $0.01 per 1M input tokens; output free. Optional web grounding $0.001 per search. Confirm on your own account.

Inputs and outputs

Recorded modality support shows what this model can accept and produce. An undocumented modality is shown as unknown rather than assumed to be unsupported.

Accepts

  • Text input
  • Image input

Produces

  • Text output

Capabilities

VisionOpen weights

What is pplx-decider-flash-v1.3?

Half the price, roughly a fifth of the parameters, same request schema. That is the whole pitch for the flash edition of Perplexity's November 2026 decider release, and for many pipelines it is enough.

Like its larger sibling, it never writes free text. It reads a state and answers typed questions with probabilities: pick an option, judge true or false, place on a scale, or locate exact spans. It is distilled from the 27B model and supports the same capabilities, including the new locate type.

Cost first

Flash v1.3 27B v1.3
Input (indicative) $0.01 per 1M tokens $0.02 per 1M tokens
Output Free Free
Alias pplx-decider-flash-latest pplx-decider-latest
Self-host memory (BF16) ~8.5 GiB + working memory ~49 GiB + working memory

At a cent per million input tokens, the bill for classifying ten million short support tickets is dominated by how long the tickets are, not by the model. Web grounding, if you enable it on the Decisions API, adds $0.001 per search regardless of which decider you choose, so on grounded workloads the price gap narrows quickly. Confirm current rates in your account.

Is it 4B or 5B?

Perplexity's API notes call it a 4B model. The published weights, built on Qwen3.5-4B, count about 5B parameters in their metadata. For billing it makes no difference; for sizing a GPU, use the ~8.5 GiB weight figure rather than either label.

Limits you will hit

  • Images are accepted and resized to at most 1.6 megapixels, which costs about 1,600 sequence positions each.
  • locate questions read text only and are capped at 128 per request.
  • Each question takes 1–255 options (2–255 for score).
  • The self-hosted default length is 32,768 tokens, and over-length inputs are rejected, never silently truncated.

The weights are released under Apache-2.0 in a repository that requires an access token.

When to pick the bigger model instead

Perplexity frames flash as the choice when latency or cost matters more than the last points of accuracy, and publishes no head-to-head accuracy figure. So make your own. Label 300 examples from the hardest slice of your data, the borderline cases, and run both IDs. If flash agrees with the 27B model on 98% of easy items but drifts on the borderline ones, a cascade works well: send everything to flash, and escalate only when its top probability falls inside an uncertainty band you choose.

Remember that thresholds tuned on the v1.2 flash model will not carry over; probabilities move between versions. For broader options, browse structured decision models or text classification.

How to evaluate pplx-decider-flash-v1.3

Compare the model on the work you need to complete.

Tasks and inputs

pplx-decider-flash-v1.3 is listed for Text classification, Zero-shot classification, Structured Decisions. It accepts text, image and produces text. Use the publisher’s input format and preprocessing for your chosen deployment.

Capacity and cost

The listed route has starting rates of $0.01 input per million tokens. Measure a completed task using the prompts, documents and tools your application will use. Include reasoning and retries in the result.

Choosing a deployment

The weights are downloadable under Apache-2.0. Check runtime support and size the complete model and working context for your expected concurrency.

Frequently asked questions

Indicatively $0.01 per 1M input tokens, half the 27B decider's rate, with free output. Confirm on your account.

These records share a directory category. Check the source card, output type and deployment requirements before treating a model as a replacement.

Google

Gemini Decision 1

Gemini Decision 1 powers Google's Decisions API, returning calibrated yes/no, choice and rubric answers instead of text, at $0.10 per 1M input tokens.

262.14K context

Perplexity

pplx-decider-v1.3

pplx-decider-v1.3 is Perplexity's 27B-class decision model: it returns probabilities for choice, yes/no, score and new span-locate questions, not text.

32.77K context

Butler

Cicero

Cicero is Butler Compute's open-weight MoE decision model: up to 64 typed questions per state, a calibrated probability for every option, no text.

131.07K context

Morph

Reflex

Morph Reflex is a family of fast classifier models that label agent turns (jailbreak, looping, frustration and more) with per-class scores.

65.54K context

Sarvam

Sarvam Prism

Sarvam Prism is a 105B decision model that returns calibrated probabilities for yes/no, choice and score questions in English and 10 Indian languages.

131.07K context

levanto

Sage 2

Levanto Sage 2 is listed as a new version of Levanto's hosted Sage decision model; Levanto's own docs do not yet name a version 2.

— context